How Many Samples to Collect

 How Many Samples to Collect

In the previous section we considered the amount of sample needed to minimize the sampling variance. Another important consideration is the number of samples required to achieve a desired maximum sampling error. If samples drawn from the target population are normally distributed, then the following equation describes the confidence interval for the sampling error


where ns is the number of samples and ss is the sampling standard deviation. Rear- ranging and substituting e for the quantity (μ– X–), gives the number of samples as

                     7.7

where ss2 and e2 are both expressed as absolute uncertainties or as relative uncertain- ties. Finding a solution to equation 7.7 is complicated by the fact that the value of t depends on ns. As shown in Example 7.8, equation 7.7 is solved iteratively.


This is not an uncommon problem. For a target population with a relative sampling variance of 50 and a desired relative sampling error of ±5%, equation 7.7 predicts that ten samples are sufficient. In a simulation in which 1000 samples of size 10 were collected, however, only 57% of the samples resulted in sampling errors of less than ±5%. By increasing the number of samples to 17 it was possible to ensure that the desired sampling error was achieved 95% of the time.

Comments

Post a Comment

Popular posts from this blog

Where to Sample the Target Population

Image
  Where to Sample the Target Population Sampling   errors   occur   when   a   sample’s   composition   is   not   identical   to   that   of   the population   from   which   it   is   drawn.   When   the   material   being   sampled   is   homoge- neous,   individual   samples   can   be   taken   without   regard   to   possible   sampling   errors. Unfortunately,   in   most   situations   the   target   population   is   heterogeneous   in   either time   or   space.   As   a   result   of   settling,   for   example,   medications   available   as   oral   sus- pensions   may   have   a   higher   concentration   of   their   active   in...
Read more

Basic Components of Spectroscopic Instrumentation

Image
  Basic Components of Spectroscopic Instrumentation The   instruments   used   in   spectroscopy   consist   of   several   common   components, including   a   source   of   energy   that   can   be   input   to   the   sample,   a   means   for   isolat- ing   a   narrow   range   of   wavelengths,   a   detector   for   measuring   the   signal,   and   a   sig- nal   processor   to   display   the   signal   in   a   form   convenient   for   the   analyst.   In   this section   we   introduce   the   basic   components   used   to   construct   spectroscopic   in  struments. Sources of Energy All   forms   of   spectroscopy   require   a ...
Read more

Quantitative Applications - Coulometric Methods of Analysis

Image
  Quantitative Applications Coulometry   may   be   used   for   the   quantitative   analysis   of   both   inorganic   and   organic compounds. Examples of controlled-potential and controlled-current   coulometric methods   are   discussed   in   the   following   sections. Controlled-Potential Coulometry  The majority of controlled-potential coulometric analyses   involve   the   determination   of   inorganic   cations   and   anions,   including   trace metals   and   halides.   Table   11.8   provides   a   summary   of   several   of   these   methods. The ability to control selectivity by carefully selecting the working electrode’s potential,   makes   controlled-potential   coulometry   particularly   useful   for   the   analy...
Read more